Dataset on a reliability generalization meta-analysis of the Oxford COVID-19 vaccine hesitancy scale

The Oxford COVID-19 Vaccine Hesitancy Scale is a 7-item psychometric scale developed by Freeman and colleagues a year after detecting the first case of the disease in 2019. The scale assesses people's thoughts, feelings, and behavior toward COVID-19 vaccines. A comprehensive search of major electronic databases, including Scopus, Clarivate Analytics, and PubMed, was conducted to extract eligible articles for inclusion in this meta-analysis. This paper reports information on data collected for a reliability generalization meta-analysis of the Oxford COVID-19 Vaccine Hesitancy Scale. The dataset incorporates information on the average reliability of the scale as measured with Cronbach's alpha in 20 studies included in the meta-analysis. Several benefits can be derived from the dataset. In particular, the research community would find this dataset beneficial as it can enhance their understanding of the health challenges of COVID-19, helping them come up with better solutions to eradicate the disease.


Value of the Data
• Researchers can use the dataset to replicate and evaluate the initial findings and explore new research questions related to the original meta-analysis, thereby contributing to the development of psychometric theory.• Researchers and public health practitioners alike would find this dataset beneficial as it can enhance their understanding of the health challenges of COVID-19, helping them come up with better solutions to eradicate the disease.
• The dataset has the potential to trigger collaborations among researchers and professionals from diverse backgrounds, allowing them to initiate new and impactful research that could benefit society.• Researchers can use the dataset cheaply and in transparent and credible ways as they do not need to repeat the entire process of generating the original data from the beginning.• University and college teachers can adopt the dataset as a teaching resource, helping students learn quantitative research with original rather than synthetic data.

Background
Researchers have developed several measurement scales to understand better and address COVID-19 vaccine hesitancy [ 1 , 2 ].Of these measurement scales, the Oxford COVID-19 Vaccine Hesitancy Scale [ 3 ] was the most widely cited by research communities [ 4 , 5 ].The initial development and validation of [ 3 ] contribute to the theory and concept development in vaccine hesitancy research.Following the initial administration of the Oxford COVID-19 Vaccine Hesitancy Scale in the UK, several studies have utilized the scale to validate it across different research contexts worldwide.Given the abundance of published works that adopted/adapted the Oxford COVID-19 Vaccine Hesitancy Scale, it is imperative to integrate the findings of these studies by conducting a meta-analysis to get valuable insights into the scale's reliability across different research settings.Performing a meta-analysis would improve our understanding of the reliability and generalizability of the Oxford COVID-19 Vaccine Hesitancy Scale across various research settings.However, the success of every meta-analysis depends mainly on the availability of the data from published and unpublished works.Thus, the purpose of this paper was to provide detailed information on data collected for a reliability generalization meta-analysis of the Oxford COVID-19 Vaccine Hesitancy Scale.

Data Description
The data frame is structured into three columns and twelve rows ranging from study identification to publication status.Before conducting the analyses, some of the variables were coded.In After creating the data frame, we organized the data into six tables and four figures.Table 1 presents the characteristics of the 20 studies included in the meta-analysis.Table 2 shows the Random-Effects Model.After presenting the heterogeneity statistics in Table 3 , we show the Influence statistics in Table 4 .The mixed-effects model with mean age, gender, study context, study quality, design of the study, and publication status as moderators is presented in Table 5 .The assessment of publication bias is provided in Table 6 .Fig. 1 depicts the Baujat plot to rule out studies that might contribute to heterogeneity.The graphical assessment outliers (i.e., influential studies) have been illustrated in Fig. 2 .Finally, Figs. 3 and 4 depict the forest and funnel plots, respectively.

Materials and Methods
A comprehensive search of electronic databases was performed to extract eligible studies for this meta-analysis, including SCOPUS, Web of Science, Google Scholar, PsycINFO, JSTOR, and PubMed.The following search terms were specifically applied: "Freeman COVID-19 vaccine hesitancy scale," OR "COVID-19 vaccine hesitancy," OR "vaccine hesitancy scale."We specifically implemented a step-by-step procedure to select eligible articles for inclusion in this meta-analysis.In step one, we found 1839 records.These records were reduced to 1096 after removing duplicates in step two.In step three, 1032 records were excluded for several reasons, such as review articles ( n = 17).Step three involves finding full-text records; our search yielded 64 full-text records in this step.However, of these records, the required statistics needed to extract data    were missing from 44 articles, and emails were sent to authors to request the required information.After sending follow-up emails to the authors and receiving no response, these 44 records were excluded in the final step.Hence, 20 articles were included in the meta-analysis [ 6 ].
We utilized two open-source software to perform the meta-analysis, namely R (version 4.3.0)and the metafor package (version 4.0 [ 7 , 8 ].Additionally, Cronbach's alpha was incorporated as the outcome measure.The result from the random-effects model shows that the model fits well with the data.We adopted the restricted maximum-likelihood estimator technique to establish the amount of heterogeneity (i.e., τ ^ 2) [ 9 ].Besides this, the Q-test for heterogeneity and the I2 statistic were also reported to understand further the heterogeneity among the studies included in the meta-analysis [ 10 , 11 ].The result suggests that regardless of the results of the Q-test, heterogeneity has been detected among the studies included in the meta-analysis ( τ ^ 2 > 0) [ 12 ].
We employed Studentized residuals and Cook's distances to rule out the possibility of outliers  or influence in the model context, as shown in Table 4 and Fig.
2 [ 8 ].Statistically, studies that turn out to have a studentized residual larger than the 100 X (1 -0.05/(2 x fc))th percentile of a standard normal distribution are deemed as potential outliers based on Bonferroni correction with two-tailed a = 0.05 for k studies included in the meta-analysis).Relatedly, we considered studies influential when a Cook's distance larger exceeds the median plus six times the interquartile range of the Cook's distances.We also utilized funnel plot, rank correlation test, as well as the regression test to determine the funnel plot asymmetry based on standard error for the observed outcomes as a predictor variable [ 13 , 14 ].

Limitations
One of the main limitations of this data is its size.We were not fortunate enough to generate relatively more comprehensive data because many authors of the previous work did not respond to our emails requesting missing statistical information from their work despite several reminders sent to them.Given that recent studies [15][16][17][18] suggest that COVID-19 vaccine hesitancy is still prevalent, it would be interesting to build on the current data by collecting more data to understand further the reliability of Oxford COVID-19 Vaccine Hesitancy scale and source of heterogeneity in the meta-analysis.

Fig. 1 .
Fig. 1.The Baujat plot.The figure provides information about how each individual study contributed to the observed heterogeneity in the meta-analysis.Each individual study included in the meta-analysis represents a point in the plot.Studies with the most significant variation from the overall effect size estimate and contributed to the observed heterogeneity are shown in the upper-left corner of the plot.

Fig. 2 .
Fig. 2. Outlier and influential study plots.The rstudent (standardised residuals), dffits (Difference in Fits), cook.d(Cook's distances), cov.r (covariance ratios), tau2.del(estimates of the amount of heterogeneity), QE.del (test statistics for heterogeneity), hat (Hat Values), and weight (Study Weights) were utilized to provide information about the diagnostic measures and influence of 20 studies examining the Reliability Generalization Meta-analysis of the Oxford COVID-19 Vaccine Hesitancy Scale.

Fig. 3 .
Fig.3.Forest plot.Forest plot displaying the alpha coefficients of individual studies with their corresponding 95 % confidence intervals.The pooled estimate of Cronbach's alpha coefficient across all studies is represented using a diamond at the bottom of the plot.

Fig. 4 .
Fig.4.Depicts the funnel plot, providing information about the link between the standard error of the effect and effect sizes, plotted on the horizontal and vertical axes, respectively.Specifically, the Funnel plot suggests that as the standard error increases, the effect sizes vary across studies, with each dot representing an individual study included in the metaanalysis.

Table
Subject Business, Management, and Decision Sciences.Specific subject area Public Health and Health Policy, Data Mining and Statistical Analysis Data format Aggregate, raw primary data (.csv), tables, and figures.Type of data Raw, coded data, screened, and partially analyzed.Data collection Data was collected from the published articles identified through a comprehensive search of electronic databases, including SCOPUS, Web of Science, Google Scholar, PsycINFO, JSTOR, and PubMed.

Table 1
Studies characteristics.

Table 6
Assessment of publication bias.
Regression Test for Funnel Plot Asymmetry Model: mixed-effects meta-regression model Predictor: standard error Test for Funnel Plot Asymmetry: z = −0.0851,p = 0.9321